ostream operators, MeshcatVisualizer fix, removal of ineffective SIMD reduction

[wxmerkt]

This PR includes:

1. operator<< for shooting problem, action model, and differential action model to enable print(ddp.problem) or print(ddp.running_models[t]) which is useful in debugging to get information on the problem
2. Removes the ineffective OpenMP simd instructions previously pointed out by Justin and also notified by the clang compiler as not optimised
3. Fixes the MeshcatVisualizer when using the latest Pinocchio devel branch (not sure when it broke)

[cmastalli]

Thank your for this work. It is quite clean and nice! I just left a few comments, and the most important point is about 'auto' keyword.

Regarding the displayers, we should create unittests. Feel free to include them in this PR (or create an issue if you don't have bandwidth now) :)

[proyan]

This looks good to me, Thanks Carlos for the review.

Regarding the CI failure, it seems to be an apt issue, perhaps the new packages aren't available fully. Could you please run again @wxmerkt ?

[wxmerkt]

I restarted Travis but robotpkg is still a bit in a weird sync state so it won't pass. As long as gepgitlab passed we should be good?

[proyan]

From my side it should be okay to merge. I'll let Carlos do the merging, since he provided the review

[proyan]

Thanks @wxmerkt for this nice work

[cmastalli]

@wxmerkt I have pushed a commit that fixes the code format (which includes c++98 compatibility). Note that the code format is hardly enforced in Travis and softly enforced in gitlab-ci. It is also needed to satisfy the 16.04 job (which it failed in gitlab-ci).

I will merge it after the current jobs finish (I will ignore Travis to avoid further delays).